Social Media: Misinformation and Content Moderation Issues for Congress - January 27, 2021 - Federation of American Scientists

 
CONTINUE READING
Social Media: Misinformation and
Content Moderation Issues for Congress

January 27, 2021

                             Congressional Research Service
                              https://crsreports.congress.gov
                                                     R46662
SUMMARY

                                                                                                         R46662
Social Media: Misinformation and
                                                                                                         January 27, 2021
Content Moderation Issues for Congress
                                                                                                         Jason A. Gallo
Social media platforms disseminate information quickly to billions of global users. The Pew              Section Research Manager
Research Center estimates that in 2019, 72% of U.S. adults used at least one social media site and
that the majority of users visited the site at least once a week.
                                                                                                         Clare Y. Cho
                                                                                                         Analyst in Industrial
Some Members of Congress are concerned about the spread of misinformation (i.e., incorrect or            Organization and Business
inaccurate information) on social media platforms and are exploring how it can be addressed by
companies that operate social media sites. Other Members are concerned that social media
operators’ content moderation practices may suppress speech. Both perspectives have focused on
Section 230 of the Communications Act of 1934 (47 U.S.C. §230), enacted as part of the
Communications Decency Act of 1996, which broadly protects operators of “interactive computer services” from liability for
publishing, removing, or restricting access to another’s content.

Social media platforms enable users to create individual profiles, form networks, produce content by posting text, images, or
videos, and interact with content by commenting on and sharing it with others. Social media operators may moderate the
content posted on their sites by allowing certain posts and not others. They prohibit users from posting content that violates
copyright law or solicits illegal activity, and some maintain policies that prohibit objectionable content (e.g., certain sexual or
violent content) or content that does not contribute to the community or service that they wish to provide. As private
companies, social media operators can determine what content is allowed on their sites, and content moderation decisions
could be protected under the First Amendment. However, operators’ content moderation practices have created unease that
these companies play an outsized role in determining what speech is allowed on their sites, with some commentators stating
that operators are infringing on users’ First Amendment rights by censoring speech .

Two features of social media platforms—the user networks and the algorithmic filtering used to manage content—can
contribute to the spread of misinformation. Users can build their own social networks, which affect the content that they see,
including the types of misinformation they may be exposed to. Most social media operators use algorithms to sort and
prioritize the content placed on their sites. These algorithms are generally built to increase user engagement, such as clicking
links or commenting on posts. In particular, social media operators that rely on advertising placed next to user-generated
content as their primary source of revenue have incentives to increase user engagement. These operators may be able to
increase their revenue by serving more ads to users and potentially charging higher fees to advertisers. Thus, algorithms may
amplify certain content, which can include misinformation, if it captures users’ attention.

The Coronavirus Disease 2019 (COVID-19) pandemic illustrates how social media platforms may contribute to the s pread of
misinformation. Part of the difficulty addressing COVID-19 misinformation is that the scientific consensus about a novel
virus, its transmission pathways, and effective mitigation measures is constantly evolving as new evidence becomes
available. During the pandemic, the amount and frequency of social media consumption increased. Information about
COVID-19 spread rapidly on social media platforms, including inaccurate and misleading information, potentially
complicating the public health response to the pandemic. Some social media operators implemented content moderation
strategies, such as tagging or removing what they considered to be misinformation, while promoting what they deemed to be
reliable sources of information, including content from recognized health authorities.

Congress has held hearings to examine the role social media platforms play in the dissemination of misinformation. Members
of Congress have introduced legislation, much of it to amend Section 230, which could affect the content moderation
practices of interactive computer services, including social media operators. In 2020, the Department of Justice also sent draft
legislation amending Section 230 to Congress. Some commentators identify potential benefits of amending Section 230,
while others have identified potential adverse consequences.

Congress may wish to consider the roles of the public and private sector in addressing misinformation, including who defines
what constitutes misinformation. If Congress determines that action to address the spread of misinformation through social
media is necessary, its options may be limited by the reality that regulation, policies, or incentives to affect one category of
information may affect others. Congress may consider the First Amendment implications of potential legislative actions. Any
effort to address this issue may have unintended legal, social, and economic consequences that may be difficult to foresee.

Congressional Research Service
Social Media: Misinformation and Content Moderation Issues for Congress

Contents
Introduction ................................................................................................................... 1
Overview of Social Media ................................................................................................ 2
    U.S. Social Media Use ............................................................................................... 4
    Content Moderation ................................................................................................... 6
Social Media Networks and Algorithms .............................................................................. 8
    Network Structure ..................................................................................................... 9
    Algorithmic Filtering and Prioritization....................................................................... 10
    Online Advertising................................................................................................... 12
Example of Misinformation and Social Media: COVID-19 Misinformation in 2020 ................ 14
Context for Congressional Consideration .......................................................................... 18
    Federal Proposals to Amend Section 230 ..................................................................... 19
    Commentary from Stakeholders on Amending Section 230 ............................................ 19
Considerations for Congress ........................................................................................... 21
    Potential Legislative Actions ..................................................................................... 22
    Concluding Thoughts ............................................................................................... 23

Figures
Figure 1. Percent of U.S. Adults Who Use at Least One Social Media Site, By Age ................... 5
Figure 2. Social Media Advertising Revenue ..................................................................... 13

Tables

Table B-1. Selected Legislation on Section 230 Introduced in the 116th Congress .................... 27
Table B-2. Selected Legislation Addressing COVID-19 Misinformation Introduced in the
  116th Congress ........................................................................................................... 30

Appendixes
Appendix A. Social Media Definitions ............................................................................. 25
Appendix B. Section 230 and COVID-19 Misinformation Legislation................................... 27

Contacts
Author Information ....................................................................................................... 31

Congressional Research Service
Social Media: Misinformation and Content Moderation Issues for Congress

Introduction
Social media platforms have become major channels for the dissemination, exchange, and
circulation of information to billions of users around the world. For years, Congress has been
concerned with the use of the internet to host, distribute, and exchange potentially illegal,
harmful, and objectionable content, including graphic sexual content, extremist content, content
that may incite violence, and foreign propaganda. Attention has often focused on social media
platforms, based on their ability to disseminate information quickly and widely and their use of
algorithms to identify and amplify content that is likely to generate high levels of user
engagement. 1
Some Members of Congress are concerned about social media dissemination of misinformation
(i.e., incorrect or inaccurate information, regardless of its origin or the intent of the individual
who disseminates it)2 and are exploring how social media platform operators can stop or slow that
dissemination via content moderation. Other Members’ interest in content moderation relates to
concerns that platform operators are moderating content that should not be restricted. Both
perspectives have focused on Section 230 of the Communications Act of 1934 (47 U.S.C. §230,
hereinafter Section 230), enacted as part of the Communications Decency Act of 1996. 3 Section
230 broadly protects interactive computer service providers,4 including social media operators,
and their users from liability for publishing, and in some instances removing or restricting access
to, another user’s content.
An example of the role social media can play in the dissemination of information and
misinformation can be seen with the Coronavirus Disease 2019 (COVID-19) pandemic. 5 The
spread of COVID-19 misinformation has complicated the public health response to COVID-19. 6

1Algorithms are computer processes that set rules for the data social media platforms receive. T hey help operators sort
and prioritize content and can be used to tailor what a user sees at a particular time. For more information, see
Appendix A.
2 Others sometimes use misinformation t o mean incorrect or inaccurate information spread by someone believing it to
be true, as distinct from disinformation, a term they reserve for false information deliberately spread to gain some
advantage. For additional information on the definitions of misinformation and disinformation, see CRS In Focus
IF11552, Considering the Source: Varieties of COVID-19 Information, by Catherine A. T heohary; Caroline Jack,
Lexicon of Lies: Terms for Problematic Information, Data & Society Research Institute, August 9, 2017, at
https://datasociety.net/pubs/oh/DataAndSociety_LexiconofLies.pdf; Claire Wardle and Hossein Derakhshan, “ T hinking
about ‘Information Disorder’: Formats of Misinformation, Disinformation, and Mal-Information,” in Cherilyn Ireton
and Julie Posetti, Journalism, Fake News & Disinformation: Handbook for Journalism Education and Training (Paris:
UNESCO Publishing, 2018), pp. 43-54, at https://en.unesco.org/sites/default/files/f._jfnd_handbook_module_2.pdf.
3
  47 U.S.C. §230. While this provision is often referred to as Section 230 of the Communications Decency Act of 1996
(P.L. 104-104), it was enacted as Section 509 of the T elecommunications Act of 1996 , which amended Section 230 of
the Communications Act of 1934. See CRS Legal Sidebar LSB10306, Liability for Content Hosts: An Overview of the
Communication Decency Act’s Section 230, by Valerie C. Brannon, and CRS Report R45650, Free Speech and the
Regulation of Social Media Content, by Valerie C. Brannon; Jeff Kosseff, The Twenty-Six Words That Created the
Internet (Ithaca, NY: Cornell University Press, 2019).
4
  47 U.S.C. §230(f)(2) defines an interactive computer service as “any information service, system, or access software
provider that provides or enables computer access by multiple users to a computer server, including specifically a
service or system that provides access to the Internet and such systems operated or services offered by libraries or
educational institutions.”
5
  For example, the World Health Organization has described the “over-abundance of information—some accurate and
some not”—that has accompanied the COVID-19 pandemic as an “infodemic.” World Health Organization, Novel
Coronavirus (2019-NCoV) Situation Report-13, February 2, 2020, at https://www.who.int/docs/default-source/
coronaviruse/situation-reports/20200202-sitrep-13-ncov-v3.pdf.
6 One proposed definition of health misinformation is information about a health phenomenon that is “contrary to the ...

Congressional Research Service                                                                                         1
Social Media: Misinformation and Content Moderation Issues for Congress

Public health communication plays a critical role in overcoming uncertainty and informing policy
and individual decisions. 7 This highlights the challenge of identifying misinformation during a
pandemic caused by a novel virus, particularly because the scientific consensus is under constant
revision and not always unanimous. It also highlights the challenge of determining the accuracy
of information in conditions of uncertainty. In some cases, misinformation may be easily
identified by the content moderators employed by social media operators as information that is
verifiably false, while in others, what is accurate or inaccurate may be a matter of judgement
based on available evidence.
This report explores the role that social media can play in the spread of misinformation—in
addition to beneficial information—using the spread of incorrect or inaccurate COVID-19
information as an example. The report provides an overview of social media and content
moderation. It focuses on three main factors that contribute to the amplification and spread of
potential misinformation on social media—(1) the use of data mining and algorithms to sort,
prioritize, recommend, and disseminate information; (2) the maximization of user engagement,
and online advertising revenue for some social media operators, as the foundation of social media
companies’ business models; and (3) the range of content moderation practices across social
media platforms. It discusses options some Members of Congress have proposed to alter
incentives surrounding social media moderation practices to address potential misinformation and
concerns that other Members have raised about censorship. The report concludes with questions
that Congress might consider as it debates whether or not to take legislative action.

Overview of Social Media
Distinguishing features of social media include the primacy of user-generated content,8 the use of
algorithms by the social media operators to sort and disseminate content, and the ability of users
to interact among themselves by forming social networks (see Appendix A for definitions of
social media sites, users, algorithms, platforms, enabling infrastructure, and operators).9 Social
media users are both the producers and consumers of content. They can post text, images, and
videos and consume others’ content by viewing, sharing, or reacting to it. 10 Users access social

consensus of the scientific community,” but with the caveat that “ what is considered true and false is constantly
changing as new evidence comes to light and as techniques and methods are advanced.” Briony Swire-T hompson and
David Lazer, “Public Health and Online Misinformation: Challenges and Recommendations,” Annual Review of Public
Health 41, no. 1 (2020), pp. 433-451, at https://doi.org/10.1146/annurev-publhealth-040119-094127.
7
  Nicole M. Krause, Isabelle Freiling, Becca Beets, et al., “Fact-Checking as Risk Communication: T he Multi-Layered
Risk of Misinformation in T imes of COVID-19,” Journal of Risk Research, April 22, 2020, pp. 1-8, at https://doi.org/
10.1080/13669877.2020.1756385.
8
  Users can be individuals, organizations, government agencies, and private firms, including news media (e.g.,
Washington Post, Fox News, New York Times).
9 Jonathan Obar and Steve Wildman, “Social Media Definition and the Governance Challenge: An Introduction to the

Special Issue,” Telecommunications Policy, vol. 39, no. 9 (2015), pp. 745-750, at https://papers.ssrn.com/sol3/
papers.cfm?abstract_id=2663153. In this report, when we refer to social media operators, we are focused primarily on
the owners of the top nine social media sites, according to a 2019 survey conducted by the P ew Research Center (Pew
Research Center, Social Media Fact Sheet, June 12, 2019, at https://www.pewresearch.org/internet/fact-sheet/social-
media/).
10
  Users can share content on social media sites by posting and reposting content or by sharing the initial post to select
individuals or to their entire network. Users can react to content by commenting on it or by “liking” it, indicating that
the user supports or “likes” the post. Some social media sites allow users to express different reactions as well. For
example, Facebook allows users to select an emoji (an icon expressing the emotion of the user), including a thumbs-up,
smiling face, frowning face, and a heart.

Congressional Research Service                                                                                          2
Social Media: Misinformation and Content Moderation Issues for Congress

media platforms through internet-based interfaces, that is, websites or mobile applications (apps).
Social media operators host user-generated content on their platforms and “organize it, make it
searchable, and [ ... ] algorithmically select some subset of it to deliver as front-page offerings,
news feeds, subscribed channels, or personalized recommendations.”11 The technical
infrastructure of social media platforms enables connections to other sites, apps, and data, and
may allow third-party developers to build applications and services that integrate with platforms,
which could provide third-parties access to some user data and preferences.
Many social media operators do not charge their users to establish accounts and use at least some
of their services. 12 These operators rely on revenue from advertisements they serve to users and
collect users’ data to target certain advertisements to specific users.13 User data includes
information about personal characteristics, preferences, and opinions provided by users when
setting up accounts, as well as information gleaned from posted content and online behaviors. The
Interactive Advertising Bureau, an industry trade association, and the research firm eMarketer
estimate that U.S. social media advertising revenue was roughly $36 billion in 2019, making up
approximately 30% of all digital advertising revenue. 14
Social media sites benefit from network effects; that is, an increasing number of users increases
the value of the site to other users. 15 For example, an individual wishing to notify multiple
acquaintances about moving to a new city may choose to share the news on a specific social
media site if his or her acquaintances also use the site. Users may have accounts with multiple
social media sites, such that increased usage of one site may reduce the amount of time the user
spends on another. Therefore, social media operators have a strong incentive to capture as much
of their users’ attention as possible. They commonly employ computational techniques to promote
content that generates strong user engagement, which can be measured by the number of clicks on
links or the amount of time spent reading posts. Some social media sites allow users to link to
content provided on other sites, permitting users to share content with larger networks and
potentially increasing traffic on the sites.
Social media operators may remove, slow the spread of, or offer warnings for content they deem
objectionable. Social media operators are broadly protected from liability for publishing, and in
some instances removing or restricting access to, another user’s content by Section 230. 16 The
authors of Section 230, former Representative Chris Cox and former Representative and current

11
  T arleton Gillespie, “Platforms Are Not Intermediaries,” Georgetown Technology Law Review, vol. 2, no. 2 (2018),
pp. 198-216, at https://georgetownlawtechreview.org/wp-content/uploads/2018/07/2.2-Gilespie-pp-198-216.pdf.
12
   Some social media operators, such as LinkedIn and Reddit, offer a premium version of their site with additional
services for a monthly fee. Others allow users (which do not include advertisers) to access all of their services without a
monthly fee (e.g., Facebook, T witter). A few operators, such as WeMe, obtain their revenue from subscription fees and
from selling custom emojis rather than online advertising.
13
   David M. Lazer, Matthew A. Baum, Yochai Benkler, et al., “T he science of fake news,” Science, vol. 359, no. 6380
(March 9, 2018), pp. 1094-1096, at https://science.sciencemag.org/content/359/6380/1094; Burt Helm, “How
Facebook’s Oracular Algorithm Determines the Fates of Start -Ups,” New York Times, November 2, 2017, at
https://www.nytimes.com/2017/11/02/magazine/how-facebooks-oracular-algorithm-determines-the-fates-of-start-
ups.html.
14 Interactive Advertising Bureau, Internet Advertising Revenue Report: Full Year 2019 Results & Q1 2020 Revenues,

May 2020, prepared by PricewaterhouseCoopers, at https://www.iab.com/wp-content/uploads/2020/05/FY19-IAB-
Internet-Ad-Revenue-Report_Final.pdf; Debra Aho Williamson, US Social Trends for 2020: eMarketer’s Predictions
for the Year Ahead, eMarketer, January 15, 2020, at https://www.emarketer.com/content/us-social-trends-for-2020.
15
  Arjun Sundararajan, “Network Effects,” author’s website, New York University Stern School of Business, at
http://oz.stern.nyu.edu/io/network.html, viewed December 23, 2020.
16   47 U.S.C. §230.

Congressional Research Service                                                                                           3
Social Media: Misinformation and Content Moderation Issues for Congress

Senator Ron Wyden, have each stated that their intent was to enable free speech and allow
interactive computer services to moderate content without government intervention. 17 Section 230
has two relevant sections regarding content hosting and moderation: Section 230(c)(1), which
states that interactive computer service providers and users may not “be treated as the publisher
or speaker of any information provided by another” person; and Section 230(c)(2), which states
that interactive computer service providers and users may not be “held liable” for any “good
faith” action “to restrict access to or availability of material that the provider or user considers to
be obscene, lewd, lascivious, filthy, excessively violent, harassing, or otherwise objectionable. ”18

U.S. Social Media Use
The Pew Research Center estimated that in 2019, 72% of U.S. adults, or about 184 million U.S.
adults, 19 used at least one social media site, based on the results of a series of surveys. 20 This was
up from 5% in 2005. Use varied by age, with the highest percentages using social media being
among the 18-29 year old and 30-49 year old cohorts (see Figure 1). Another report estimates
that in January 2020, there were roughly 230 million social media users in the United States of all
ages (13 is a standard minimum age to register an account on many social media sites), and that
users subscribed to an average of roughly seven social media accounts. 21 The majority of U.S.
social media users report visiting the sites weekly and many report visiting the sites daily. 22

17
   T estimony of Christopher Cox in U.S. Congress, Senate Committee on Commerce, Science, and T ransportation,
Communications, T echnology, Innovation, and the Internet, The PACT Act and Section 230: The Impact of the Law
that Helped Create the Internet and an Examination of Proposed Reforms for Today’s Online World , 116 th Cong., 2 nd
sess., July 28, 2020, at https://www.commerce.senate.gov/services/files/BD6A508B-E95C-4659-8E6D-
106CDE546D71; Christopher Cox, “Policing the Internet: A Bad Idea in 1996 –and T oday,” RealClear Politics, June
25, 2020, at https://www.realclearpolitics.com/articles/2020/06/25/policing_the_internet_a_bad_idea_in_1996_—
_and_today.html; Ron Wyden, “I wrote this law to protect free speech. Now T rump wants to revoke it,” CNN Business
Perspectives, June 9, 2020 at https://www.cnn.com/2020/06/09/perspectives/ron-wyden-section-230/index.html.
18 CRS Legal Sidebar LSB10484, UPDATE: Section 230 and the Executive Order on Preventing Online Censorship ,

by Valerie C. Brannon et al.
19 CRS analysts calculated the 184 million U.S. adult figure using U.S. Census Bureau population estimates. T he
Census Bureau estimates that on July 1, 2019, there were 328,239,523 people in the United States and that 77.7% of
these were 18 years or older. Census Bureau, QuickFacts: United States, at https://www.census.gov/quickfacts/fact/
table/US/PST 045219.
20 Pew Research Center, Social Media Fact Sheet, June 12, 2019, at https://www.pewresearch.org/internet/fact-sheet/

social-media/.
21
  Simon Kemp, Digital 2020: The United States of America, Datareportal, February 11 2020, slide 17 and 42, at
https://datareportal.com/reports/digital-2020-united-states-of-america.
22Social Media Fact Sheet. T he Pew Research Center survey results indicate that 74% of Facebook, 63% of Instagram,
61% of Snapchat, 51% of YouT ube, and 42% of T witter users report daily use in 2019.

Congressional Research Service                                                                                        4
Social Media: Misinformation and Content Moderation Issues for Congress

     Figure 1. Percent of U.S. Adults Who Use at Least One Social Media Site, By Age

      Source: Pew Research Center, Internet and Technology, “Social Media Fact Sheet,” June 12, 2019.
      https://www.pewresearch.org/internet/fact-sheet/social-media/. Based on surveys conducted 2005-2019.

Media consumption, including social media use, has increased during the COVID-19 pandemic.
This is likely a result, primarily, of entertainment venue closures and an increased amount of time
spent at home as many employees and students shifted to remote work and school. The Nielsen
Company reported a 215% increase in time spent on mobile devices accessing current news in the
United States in March 2020 compared to the year before. 23 Facebook reported an increase of
over 50% in total messaging across its offerings globally from February 2020, before most
countries in Europe and North America had closed schools, offices, and public venues, to March
2020, when shutdowns became widespread.24 In April 2020, Kantar, a data and market research
firm that surveyed over 25,000 individuals in 30 global markets, reported that social media usage
had increased globally by 61% over normal usage rates since the start of the COVID-19
pandemic. 25
Social media sites also serve as major venues for the circulation of digital content from both
online-only and traditional print and broadcast news outlets. 26 Prior to the COVID-19 pandemic, a
2019 Pew Research Center report found that 55% of surveyed U.S. adults reported accessing
news through social media sites, and that 52% of U.S. adults reported using Facebook to access
news. 27 The report also states that 88% of U.S. adults were aware that social media operators
23T he Nielson Company, “COVID-19: T racking the Impact on Media Consumption,” June 16, 2020, at
https://www.nielsen.com/us/en/insights/article/2020/covid-19-tracking-the-impact-on-media-consumption.
24
   Alex Schultz and Jay Parikh, “Keeping Our Services Stable and Reliable During the COVID-19 Outbreak,” About
Facebook, March 24, 2020, at https://about.fb.com/news/2020/03/keeping-our-apps-stable-during-covid-19/.
25
   Kantar, “COVID-19 Barometer: Consumer Attitudes, Media Habits and Expectations,” April 3, 2020 , at
https://www.kantar.com/inspiration/coronavirus/covid-19-barometer-consumer-attitudes-media-habits-and-
expectations.
26
   Philip M. Napoli, Social Media and the Public Interest: Media Regulation in the Disinformation Age (New York:
Columbia University Press, 2019), pp. 1-2.
27 Elisa Shearer and Elizabeth Grieco, Americans Are Wary of the Role Social Media Sites Play in Delivering the News,

Pew Research Center, October 2, 2019, at https://www.journalism.org/2019/10/02/americans-are-wary-of-the-role-

Congressional Research Service                                                                                      5
Social Media: Misinformation and Content Moderation Issues for Congress

exert some control over the mix of news that users see on their sites, and that 62% believe that
these operators have too much control over news content. 28

Content Moderation
Social media operators maintain policies that prohibit users from posting certain content, such as
content that exhibits graphic violence, child sexual exploitation, and hateful content or speech. 29
An operator may temporarily or permanently ban users that violate its policies, depending on the
operator’s perspective on the severity of the users’ violation(s). There is no uniform standard for
content moderation, resulting in practices varying across social media sites. 30 Some operators
have chosen to release reports containing information on their content moderation practices, such
as the amount of content removed and the number of appeals, 31 but operators are not required to
release this information.
Social media operators rely on several sources to identify content to flag or remove: (1) users, (2)
content moderators, and (3) automated systems, also known as artificial intelligence (AI)
technologies. 32 Users can flag or mark inappropriate posts for content moderators to review and
remove when applicable. Automated systems can also flag and remove posts. Content
moderators, primarily contractors, may be able to identify nuanced violations of content policy,
such as taking into account the context of a statement.33 For example, in the first quarter of 2020,
AI technology flagged 99% of violent and graphic content and child nudity on Facebook for
review before any user reported it. 34 In contrast, Facebook’s AI technology identified only 16% of
bullying and harassment content, suggesting content moderators are better able to identify this
form of policy violation.
Some social media operators may be compelled to rely more heavily on AI technologies to
moderate content. Some commentators have raised concern about whether repeatedly reviewing
graphic, explicit, and violent materials harms content moderators’ mental health. 35 For example,
in 2020, Facebook reached a settlement in a class-action lawsuit filed by its content moderators

social-media-sites-play-in-delivering-the-news/.
28
  Ibid.
29
  For example, Facebook and T witter provide lists of inappropriate content at https://www.facebook.com/
communitystandards/introduction and https://help.twitter.com/en/rules-and-policies/twitter-rules, respectively.
30
   Marietje Schaake and Rob Reich, Election 2020: Content Moderation and Accountability, Stanford University
Human-Centered Artificial Intelligence, Stanford Freeman Spogli Institute Cyber Policy Center, Issue Brief, October
2020, at https://hai.stanford.edu/sites/default/files/2020-10/HAI_CyberPolicy_IssueBrief_3.pdf.
31 For example, the latest reports released by Facebook and T witter are available at https://transparency.facebook.com/

community-standards-enforcement and https://transparency.twitter.com/en/reports/removal-requests.html, respectively.
32
   T arleton Gillespie, “Content Moderation, AI, and the Question of Scale,” Big Data & Society, vol. 7, no. 2, (2020):
pp. 1-5, at https://doi.org/10.1177/2053951720943234. According to the article, only a few operators use machine
learning techniques to identify new content that violates the social media sites’ policies. Most operators rely primarily
on algorithms that are coded to identify specific phrases and images.
33
  For example, according to a class-action lawsuit filed in September 2018 against Facebook and Pro Unlimited,
Facebook had content moderators review more than 10 million potentially rule-breaking posts per week and sought to
review all user-reported violations within 24 hours (Selena Scola v. Facebook Inc. and Pro Unlimited Inc., 18 CIV
05135 (San Mateo County Superior Court), at https://assets.documentcloud.org/documents/6889335/18-CIV-05135-
Complaint.pdf). Social media operators do not publicly disclose the number of content violations that are flagged by
users, content moderators, and AI technologies.
34
  Paul Barrett, “Who Moderates the Social Media Giants? A Call to End Outsourcing,” NYU Stern Center for Business
and Human Rights, June 4, 2020, at https://bhr.stern.nyu.edu/blogs/2020/6/4/who-moderates-the-social-media-giants.
35   Ibid.

Congressional Research Service                                                                                              6
Social Media: Misinformation and Content Moderation Issues for Congress

who claimed to have experienced post-traumatic stress disorder from reviewing content on its
sites; Facebook agreed to pay $52 million to its content moderators.36 During the COVID-19
pandemic, some content moderators worked remotely, but privacy and security concerns meant
some of the content moderation was done by automated systems.37 These systems can quickly
review large volumes of content “when scale problems make manual curation or intervention
unfeasible.”38
By relying more heavily on automated systems, social media operators may mistakenly remove or
fail to remove content. Thus, some operators have stated that no account would be permanently
suspended solely by an automated enforcement system during the COVID-19 pandemic. 39 For
example, Facebook’s automated systems have reportedly removed ads from small businesses,
mistakenly identifying them as content that violates its policies and causing the businesses to lose
money during the appeals process. 40 A wide range of small businesses have reportedly been
affected by these mistakes, including a seed company for sharing a photo of Walla Walla onions
as being overtly sexual and a solar roof company that used acronyms that are similar to
cryptocurrency tokens.41 In 2019, Facebook restored 23% of the 76 million appeals it received,
and restored an additional 284 million pieces of content without an appeal—about 2% of the
content that it took action on for violating its policies. 42 During the COVID-19 pandemic, the
amount of content removed by Facebook and the amount restored without an appeal increased for
some categories—such as hate speech, bullying, and harassment—and decreased for other
categories, such as adult nudity and sexual activity. 43
Some social media operators have altered their content moderation practices over time. For
example, in 2019, Twitter and Instagram released new policies to reduce bullying and hate speech
on their sites. 44 Some of these changes may have partially been in response to criticism social
media operators received for allowing certain content on their sites, such as hate speech against
Rohingya Muslims in Myanmar that spread on Facebook. 45 Some operators have reportedly

36 Bobby Allyn, “In Settlement, Facebook to Pay $52 Million to Content Moderators with PT SD,” NPR, May 12, 2020,
at https://www.npr.org/2020/05/12/854998616/in-settlement-facebook-to-pay-52-million-to-content-moderators-with-
ptsd.
37
  Shannon Bond, “Facebook, YouT ube Warn of More Mistakes As Machines Replace Moderators,” NPR, March 31,
2020, at https://www.npr.org/2020/03/31/820174744/facebook-youtube-warn-of-more-mistakes-as-machines-replace-
moderators; Elizabeth Dwoskin and Nitasha T iku, “Facebook Sent Home T housands of Human Moderators Due to
Coronavirus. Now the Algorithms Are In Charge,” Washington Post, March 24, 2020, at
https://www.washingtonpost.com/technology/2020/03/23/facebook-moderators-coronavirus/.
38 Robert Gorwa, Reuben Binns, and Christian Katzenbach, “Algorithmic Content Moderation: T echnical and Political
Challenges in the Automation of Platform Governance,” Big Data & Society, vol. 1, no. 15 (January-June 2020), p. 3,
at https://journals.sagepub.com/doi/10.1177/2053951719897945.
39
  Shannon Bond, “Facebook, YouT ube Warn of More Mistakes As Machines Repla ce Moderators,” NPR, March 31,
2020, at https://www.npr.org/2020/03/31/820174744/facebook-youtube-warn-of-more-mistakes-as-machines-replace-
moderators; “An Update On Our Continuity Strategy During COVID-19,” Twitter, updated April 1, 2020, at
https://blog.twitter.com/en_us/topics/company/2020/An-update-on-our-continuity-strategy-during-COVID-19.html.
40
  Sarah Frier, “Facebook’s AI Mistakenly Bans Ads for Struggling Businesses,” Bloomberg, November 27, 2020, at
https://www.bloomberg.com/news/articles/2020-11-27/facebook-s-ai-mistakenly-bans-ads-for-struggling-businesses.
41
  Ibid.
42
  “Community Standards Enforcement Report,” Facebook, November 2020, at https://transparency.facebook.com/
community-standards-enforcement.
43
     T hese trends are based on comparing the numbers listed for the first and third quarters of 2020. Ibid.
44 Sara Harrison, “T witter and Instagram Unveil New Ways to Combat Hate—Again,” Wired, July 11, 2019, at
https://www.wired.com/story/twitter-instagram-unveil-new-ways-combat-hate-again/.
45 T om Miles, “U.N. Investigation Cite Facebook Role in Myanmar Crisis,” Reuters, March 12, 2018, at

Congressional Research Service                                                                                    7
Social Media: Misinformation and Content Moderation Issues for Congress

reconsidered their approach to trade-offs between free expression and safety, such as taking a
harder line with removing misinformation. 46 For example, Facebook partners with third-party
fact-checkers to review and rate the accuracy of articles and posts, placing those identified as
false lower in users’ news feeds. 47 In addition, Facebook includes information about the publisher
of articles posted on its site and displays articles from the third-party fact-checkers below posts
on the same topic. Twitter labels content containing misleading information or disputed claims
that it determines to be “moderately harmful,” while removing misleading content that it
determines to be “severely harmful.”48 These actions were taken voluntarily by Facebook and
Twitter. Currently, the decision to moderate, or to not moderate, certain content is at the discretion
of each operator.
Misinformation can spread on social media sites, even with content moderation techniques
implemented by operators. Misinformation can spread before moderators discover, review, and
remove the content. To add further complication, users can share content across social media
platforms, meaning content can spread on another platform even after the original content is
removed. Users who recontextualize the original problematic content, for example, through
reposting content or posting screenshots of it, may complicate an operator’s enforcement of its
policies. In addition, some operators may choose not to remove some content that violates its
policies. For example, Facebook’s CEO Mark Zuckerberg stated on a post, “A handful of times a
year, we leave up content that would otherwise violate our policies if the public interest value
outweighs the risk of harm.”49
Through their content moderation practices, social media operators may remove content that
some users find valuable. Some commentators and legislators have raised concern that these
operators are removing too much content, including content from whistleblowers.50 As social
media sites have grown in popularity, they have created some unease that companies determine
what speech is acceptable. 51 However, as private companies, social media operators are generally
able to determine what content is allowed on their sites. 52

Social Media Networks and Algorithms
Social media platforms are shaped by the structures of their user networks and computational
tools, such as algorithmic filtering, that operators use to manage large volumes of user-generated

https://www.reuters.com/article/us-myanmar-rohingya-facebook/u-n-investigators-cite-facebook-role-in-myanmar-
crisis-idUSKCN1GO2PN.
46
   “Social Media’s Struggle with Self-Censorship,” The Economist, October 22, 2020, at https://www.economist.com/
briefing/2020/10/22/social-medias-struggle-with-self-censorship.
47
   T essa Lyons, “Hard Questions: What’s Facebook’s Strategy for Stopping False News?,” Facebook, May 23, 2018, at
https://about.fb.com/news/2018/05/hard-questions-false-news/.
48 Yoel Roth and Nick Pickles, “Updating Our Approach to Misleading Information,” Twitter, May 11, 2020, at

https://blog.twitter.com/en_us/topics/product/2020/updating-our-approach-to-misleading-information.html.
49
   Mark Zuckerberg, Facebook, June 26, 2020, at https://www.facebook.com/zuck/posts/10112048980882521.
50 “Social Media’s Struggle with Self-Censorship,” The Economist, October 22, 2020, at https://www.economist.com/
briefing/2020/10/22/social-medias-struggle-with-self-censorship.
51
   Zeynep T ufecki, “T witter Has Officially Replaced the T own Square,” Wired, December 27, 2017, at
https://www.wired.com/story/twitter-has-officially-replaced-the-town-square/; “Social Media’s Struggle with Self-
Censorship,” The Economist, October 22, 2020, at https://www.economist.com/briefing/2020/10/22/social-medias-
struggle-with-self-censorship.
52   CRS Report R45650, Free Speech and the Regulation of Social Media Content, by Valerie C. Brannon.

Congressional Research Service                                                                                      8
Social Media: Misinformation and Content Moderation Issues for Congress

content continually posted on their sites and increase user engagement with content.53 Both
network structure and computational tools are intended to increase the number of users and user
engagement. 54 These components allow operators to increase their revenue, particularly for those
that have online advertisements on their platforms, but may also increase the spread of
misinformation that increases user engagement. Each social media operator balances incentives to
moderate and prioritize content to increase user engagement and its revenue.

Network Structure
Social media users can establish connections to other users of a site, creating social networks or
communities that can be based on common interests, relationships that exist offline, employment,
or other factors. The structure of these networks affect how individuals search for one another and
how connections are initiated and established, 55 which can also depend on the level of privacy
offered by the operator and chosen by each user. For example, some social media sites allow
users to choose whether to make their profiles open to the public or only to those who have
established connections by mutual consent.
On some social media sites, users can limit the content that they see through the networks they
choose to build. Each user can choose to follow or stop following other users, including those
who post content that the user disagrees with. Thus, social media sites can facilitate “echo
chambers” or “filter bubbles,” where a user’s ideas are reiterated and reinforced by others while
other ideas may be excluded. 56 Some research has shown that the overlap of networks (i.e., those
with common followers) increases the likelihood that two users will share content through the
network, although this effect depends on the novelty of the content. 57 Echo chambers can enhance
the spread of information, including but not limited to misinformation, particularly before the
information “goes viral” (i.e., spreads rapidly on the internet). 58
Social media operators often have economic incentives to encourage users to expand their
networks, as the value of a site to a user increases as more users join or increase their activity on

53Michael Bosetta, “T he Digital Architectures of Social Media: Comparing Political Campaigning on Facebook,
T witter, Instagram, and Snapchat in the 2016 U.S. Election,” Journalism & Mass Communication Quarterly, vol. 95,
no. 2 (2018), pp. 471-496. T he article includes datafication as a separate category and includes a fourth component—
functionality—which includes aspects such as the graphical interface. While these may be important aspects of the
social media structure, it is less relevant to the spread of misinformation, and thus not discussed in this report.
54 Renee DiResta, “Free Speech Is Not the Same As Free Reach,” Wired, August 30, 2018, at https://www.wired.com/
story/free-speech-is-not-the-same-as-free-reach/.
55
   Michael Bosetta, “T he Digital Architectures of Social Media: Comparing Political Campaigning on Facebook,
T witter, Instagram, and Snapchat in the 2016 U.S. Election,” Journalism & Mass Communication Quarterly, vol. 95,
no. 2 (2018), pp. 471-496; Danah Boyd, “Social Network Sites as Networked Publics: Affordances, Dynamics, and
Implications,” A Networked Self: Identity, Community, and Culture on Social Network Sites (New York, NY:
Routledge, 2011.)
56
   “Digital Media Literacy—What Is an Echo Chamber?,” Goodwill Community Foundation Inc. at
https://edu.gcfglobal.org/en/digital-media-literacy/how-filter-bubbles-isolate-you/1/; Christopher Seneca, “How to
Break Out of Your Social Media Echo Chamber,” Wired, September 17, 2020, at https://www.wired.com/story/
facebook-twitter-echo-chamber-confirmation-bias/.
57
   Jing Peng, Ashish Agarwal, Kartik Hosanagar, et al., “Network Overlap and Content Sharing on Social Media
Platforms,” Journal of Marketing Research, vol. 55 (August 2018), pp. 571-585, at https://journals.sagepub.com/doi/
10.1509/jmr.14.0643.
58
  Petter T örnberg, “Echo Chambers and Viral Misinformation: Modeling Fake News as Complex Contagion,” PLOS
ONE, vol. 13, no. 9 (2018); Michela Del Vicario, Alessandro Bessi, Fabiana Zollo, et al., “T he Spreading of
Misinformation Online,” Proceedings of the National Academy of Sciences, vol. 113, no. 3 (January 19, 2016), pp. 554-
559, at https://www.pnas.org/content/113/3/554.

Congressional Research Service                                                                                          9
Social Media: Misinformation and Content Moderation Issues for Congress

the site. Some social media sites recommend connections based on peripheral connections (i.e.,
someone who is a friend of one of the user’s friends) and often allow users to search for others,
using their name, email address, occupation, or other information. 59 Expanding the number of
users increases the number of possible connections and recommendations, which can encourage
even more individuals to join, exposing more users to advertisements that generate revenue for
the social media operator.

Algorithmic Filtering and Prioritization
Social media sites contain large amounts of content. Over the last decade, decreased costs of
social media enabling infrastructure have made it possible for operators to increase the amount of
user-generated content that they maintain. 60 Operators use algorithms to sort, index, curate, and
prioritize user content, as well as to suppress illegal and other content the operator chooses to
moderate. Social media operators can change or refine their algorithms to meet evolving business
goals in response to internal incentives (e.g., maximizing engagement, increasing advertising
revenue) and external pressures (e.g., user complaints, stakeholders), affecting what users see,
what content is privileged and promoted, and what content rapidly spreads across the platform
(i.e., “goes viral”). 61 Specifics about the algorithms that social media operators use are considered
proprietary and are not publicly available, although there is a general understanding of how these
algorithms work.
Each user’s activities are quantified and used to determine the selection, sequence, and visibility
of posts. 62 For example, Facebook states that its News Feed prioritizes recent content that is found
to be relevant to the user, based on factors such as previous engagement with the content
provider. 63 The algorithms also may prioritize content that is likely to sustain user engagement—
such as sharing, commenting on, or reacting to content—rather than the content’s veracity. 64
According to a Wall Street Journal article, slides presented by an internal Facebook team to
company executives in 2018 stated, “Our algorithms exploit the human brain’s attraction to
divisiveness,” and warned that the algorithms would promote “more and more divisive content in

59Danah Boyd, “Social Network Sites as Networked Publics: Affordances, Dynamics, and Implications,” in A
Networked Self: Identity, Community, and Culture on Social Network Sites (New York, NY: Routledge, 2011).
60 A definition of “social media enabling infrastructure” is provided in Appendix A. Jonathan Obar and Steve
Wildman, “Social Media Definition and the Governance Challenge: An Introduction to the Special Issue,”
Telecommunications Policy, vol. 39, no. 9 (2015), pp. 745-750, at https://papers.ssrn.com/sol3/papers.cfm?abstract_id=
2663153.
61 Leo Mirani, “ T he World in 2019: Slow Social,” The Economist, 2019, at https://worldin2019.economist.com/

slowingdownsocialmedia.
62
   T aina Bucher, “Want to be on the T op? Algorithmic Power and the T hreat of Invisibility on Facebook,” New Media
& Society, vol. 14, no. 7 (2012): 1164-1180, at https://journals.sagepub.com/doi/abs/10.1177/1461444812440159;
Michael Bosetta, “T he Digital Architectures of Social Media: Comparing Political Campaigning on Facebook, T witter,
Instagram, and Snapchat in the 2016 U.S. Election,” Journalism & Mass Communication Quarterly, vol. 95, no. 2
(2018), pp. 471-496, at https://journals.sagepub.com/doi/full/10.1177/1077699018763307.
63
   “How News Feed Works,” Facebook Help Center, accessed on October 28, 2020, at https://www.facebook.com/
help/1155510281178725; Kelley Cotter, Janghee Cho, and Emilee Rader, “Explaining the News Feed Algorithm: An
Analysis of the ‘News Feed FYI’ Blog,” In Proceedings of the 2017 CHI Conference Extended Abstracts on Human
Factors in Computing Systems - CHI EA ’17, pp. 1553-1560, Denver, CO: ACM Press, 2017, at https://doi.org/
10.1145/3027063.3053114.
64Michael Bosetta, “T he Digital Architectures of Social Media: Comparing Political Campaigning on Facebook,
T witter, Instagram, and Snapchat in the 2016 U.S. Election,” Journalism & Mass Communication Quarterly, vol. 95,
no. 2 (2018), pp. 471-496, at https://journals.sagepub.com/doi/pdf/10.1177/1077699018763307.

Congressional Research Service                                                                                     10
Social Media: Misinformation and Content Moderation Issues for Congress

an effort to gain user attention and increase time on the platform.”65 One study found that users
are more likely to read and share emotional news content and content that provides relevant and
practical information, particularly positive news. 66
Some social media operators have made changes to their algorithms. For example, in 2018,
Facebook started prioritizing “meaningful posts,” or those shared by family and friends rather
than news organizations and brands. 67 Some social media operators allow users to personalize
which content is prioritized. On May 31, 2019, Facebook launched the tool “Why am I seeing this
post?” It allows users to see why the content was posted on their news feed—such as whether the
post is from someone in their network or if they have previously interacted with similar posts—
and allows users to adjust their preferences, such as prioritizing posts from specific people or
pages. 68 On Twitter, users can prioritize content through their searches or lists they have created, 69
and can opt to see content in reverse chronological order only from accounts that a user follows.
Users can also choose to follow certain “topics,” which allows users to follow the most popular
conversations about a specific topic. 70 Information on how these changes are incorporated into the
algorithms is not publicly available.
Users can try to use the algorithms on social media to make their content go viral. They can
partner with “influencers,” that is, users with a large number of followers, 71 and try to have their
content reposted by popular accounts.72 Social media sites also benefit from content going viral,
which could attract more users and encourage users to spend more time on their sites.
Internet bots—software applications that can perform automated tasks such as rapidly posting,
liking, and recirculating content on social media sites using inauthentic accounts —can affect the
prioritization of content on social media sites and may be used to spread misinformation. 73 Bots
can post or amplify content by engaging with it. For example, a bot may be programmed to
search for and respond to posts containing specific words or phrases. This can cause algorithms

65Jeff Horowitz and Deepa Seetharaman, “Facebook Executives Shut Down Efforts to Make the Site Less Divisive,”
Wall Street Journal, May 26, 2020, at https://www.wsj.com/articles/facebook-knows-it-encourages-division-top-
executives-nixed-solutions-11590507499.
66
   Ahmed Al-Rawi, “Viral News on Social Media,” Digital Journalism , vol. 7, no. 1 (2019), pp. 63-79, at
https://www.tandfonline.com/doi/full/10.1080/21670811.2017.1387062.
67
   Hayley T sukayama, “Facebook’s Changing Its News Feed. How Will It Affect What You See?,” Washington Post,
January 12, 2018, at https://www.washingtonpost.com/news/the-switch/wp/2018/01/12/facebooks-changing-its-news-
feed-how-will-it-affect-what-you-see/; Jonah Bromwich and Matthew Haag, “Facebook Is Changing. What Does T hat
Mean to Your News Feed?,” New York Times, January 12, 2018, at https://www.nytimes.com/2018/01/12/technology/
facebook-news-feed-changes.html.
68 Ramya Sethurman, “Why Am I Seeing T his? We Have An Answer For You,” Facebook Newsroom , March 31, 2019,
at https://about.fb.com/news/2019/03/why-am-i-seeing-this/.
69 “About Your T witter T imeline,” Twitter, access on September 29, 2020 at https://help.twitter.com/en/using-twitter/

twitter-timeline.
70 “Introducing T opics,” Twitter, November 11, 2019, at https://blog.twitter.com/en_us/topics/product/2019/
introducing-topics.html.
71
   Some influencers may be paid by advertisers or users. Arielle Pardes, “Instagram Will (Finally) Pay Influencers,”
Wired, May 27, 2020, at https://www.wired.com/story/instagram-finally-pay-influencers-badges-igtv-ads/.
72
   Steve Olenski, “7 Ways to Up Your Chances of Going Viral on Social Media,” Forbes, February 6, 2018, at
https://www.forbes.com/sites/steveolenski/2018/02/06/7-ways-to-up-your-chances-of-going-viral-on-social-media.
73
   Fake, or inauthentic, accounts are profiles impersonating other individuals or organizations. An internet bot is
software that runs automated computer programs over the internet, generally capable of performing simple, repetitive
tasks faster than an individual can. Some websites use a “Completely Automated Public T uring to tell Com puters and
Humans Apart,” or CAPT CHA test, to try to identify internet bots. More information on CAPT CHA tests is available
at https://www.cloudflare.com/learning/bots/how-captchas-work/.

Congressional Research Service                                                                                         11
Social Media: Misinformation and Content Moderation Issues for Congress

used by social media operators to inadvertently prioritize misinformation. Users and social media
operators can recognize some internet bots based on various factors, such as the syntax used, the
user’s profile, or abnormal account activity. 74 Some users may choose not to engage with this
content by not sharing or reposting it, and some social media operators remove this content.
However, bots are becoming more sophisticated, making it more difficult for users and content
moderators to recognize them, particularly if a post has already gone viral. Users may
inadvertently share or like content created or shared by an internet bot. 75 Studies have indicated
that bots can contribute to the long-term spread of misinformation. 76

Online Advertising
Social media operators have economic incentives to increase user engagement on their sites,
particularly operators that rely on online advertising revenue. These operators can increase their
revenue by amplifying content that is more likely to be shared and commented on, which could
include misinformation. As a user spends more time scrolling through posts or newsfeeds, social
media operators can expose that user to more advertisements and collect more data about the user.
This increases the likelihood that the user will click on at least one advertisement and allows
operators to build better profiles of the user’s characteristics and revealed preferences. These
advertisements are often displayed as posts, generally distinguishable through labels such as
“sponsored.”
Advertising sales are the primary source of revenue for most social media operators. In 2019,
online advertising globally provided about 98% ($70 billion) of Facebook Inc.’s annual
revenue, 77 84% ($135 billion) of Google’s, 78 and 87% ($3 billion) of Twitter’s. 79 Facebook CEO
Mark Zuckerberg highlighted the importance of advertising in prepared remarks to the House
Committee on the Judiciary, Subcommittee on Antitrust, Commercial, and Administrative Law,
stating, “Facebook supports its mission of connecting people around the world by selling ads.”80
According to an Interactive Advertising Revenue report, revenue from advertising on social
media in the United States increased from about $2.9 billion in 2012 to $35.6 billion in 2019
(Figure 2), and is projected to continue increasing. 81 Based on this data, social media made up

74
  Will Knight, “How to T ell if You’re T alking to a Bot,” MIT Technology Review, July 18, 2018, at
https://www.technologyreview.com/2018/07/18/141414/how-to-tell-if-youre-talking-to-a-bot/; Ryan Detert, “Bot or
Not: Seven Ways to Detect an Online Bot,” Forbes, August 6, 2018, at https://www.forbes.com/sites/
forbesagencycouncil/2018/08/06/bot-or-not-seven-ways-to-detect-an-online-bot/.
75Kate Starbird, “Disinformation’s Spread: Bots, T rolls and All of Us,” Nature, vol. 571 (July 25, 2019), p. 449, at
https://www.nature.com/articles/d41586-019-02235-x.
76 Marina Azzimonti and Marcos Fernandes, “Social Media Networks, Fake News, and Polarization,” NBER Working
Paper 24462, March 2018, at http://www.nber.org/papers/w24462.
77 Facebook Inc. SEC Form 10-K for the year ending December 31, 2019, p. 56.

78T he percentage is calculated by dividing Google advertising revenue by Google’s revenues, not Alphabet’s
(Google’s parent company) total revenues. Alphabet Inc. SEC Form 10 -K for the year ending December 31, 2019, p.
29. YouT ube, owned by Google, generated roughly $15 billion in revenue.
79
  T witter Inc. SEC Form 10-K for the year ending December 31, 2019, p. 39.
80
  T estimony of Mark Zuckerberg, Chief Executive Officer, Facebook Inc. in U.S. Congress, House Committee on the
Judiciary, Subcommittee on Antitrust, Commercial, and Administrative Law, Online Platforms and Market Power,
Part 6: Examining the Dominance of Amazon, Apple, Facebook, and Google , hearings, 116 th Cong., 2 nd sess., July 28,
2020.
81Interactive Advertising Bureau, “Internet Advertising Revenue Report: Full Year 2019 Results & Q1 2020
Revenues,” prepared by PricewaterhouseCoopers, May 2020, pp. 19, at https://www.iab.com/wp-content/uploads/2020/
05/FY19-IAB-Internet-Ad-Revenue-Report_Final.pdf.

Congressional Research Service                                                                                          12
You can also read